starred/orca
1
0
Fork 0
mirror of https://github.com/stablyai/orca.git synced 2026-09-24 00:02:24 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(cloud): compare the Asia topology budget gate against the measured 500-connection default (#22386)

Browse Source 
The topology workflow's Cloud SQL gate carried a hard-coded 400 for the
instance's tier default while the consumer contract records the value
measured on the live instance (SHOW max_connections = 500, 2026-09-16,
#21163). The gate compares the two and the first production plan run
(35815654836) failed silently on that mismatch before Terraform ran.

The verified default now lives beside the tier and version it is verified
for, as VERIFIED_DEFAULT_MAX_CONNECTIONS, so the contract and the workflow
are two independent records of the same measurement and the gate keeps
its cross-check. The test pins the new source and forbids a bare literal.

Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
This commit is contained in:
Jinwoo Hong
2026-09-23 00:28:44 -04:00
committed by GitHub
parent 1dcbd4e65d
commit 483fa0aca2
3 changed files with 10 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.github/workflows/cloud-deploy-relay-asia-topology.yml
+5 -3
View File
@@ -55,6 +55,8 @@ jobs:
CLOUD_SQL_INSTANCE: ${{ inputs.environment == 'production' && 'orca-cloud-auth-db' || 'orca-cloud-staging-auth-db' }}
VERIFIED_DEFAULT_MAX_CONNECTIONS_TIER: db-custom-4-15360
VERIFIED_DEFAULT_MAX_CONNECTIONS_DATABASE_VERSION: POSTGRES_17
# SHOW max_connections on the live instance, 2026-09-16; no flag is set.
VERIFIED_DEFAULT_MAX_CONNECTIONS: '500'
TF_BACKEND: ${{ inputs.environment == 'production' && 'backend/production.hcl' || 'backend/staging.hcl' }}
TF_VARS: ${{ inputs.environment == 'production' && 'environments/production.tfvars' || 'environments/staging.tfvars' }}
TOPOLOGY_WORKLOAD_IDENTITY_PROVIDER: ${{ inputs.environment == 'production' && vars.PRODUCTION_GCP_RELAY_ASIA_TOPOLOGY_WORKLOAD_IDENTITY_PROVIDER || vars.STAGING_GCP_RELAY_ASIA_TOPOLOGY_WORKLOAD_IDENTITY_PROVIDER }}
@@ -119,13 +121,13 @@ jobs:
live_max="${live_flag}"
live_source=explicit-flag
else
# The verified production database uses Cloud SQL's 400-connection
# default for this exact shape; fail closed if its shape changes.
# No flag: the ceiling is the tier default measured for this exact
# shape; fail closed if the shape changes.
test "$(jq -er '.settings.tier' <<< "${instance}")" = \
"${VERIFIED_DEFAULT_MAX_CONNECTIONS_TIER}"
test "$(jq -er '.databaseVersion' <<< "${instance}")" = \
"${VERIFIED_DEFAULT_MAX_CONNECTIONS_DATABASE_VERSION}"
live_max=400
live_max="${VERIFIED_DEFAULT_MAX_CONNECTIONS}"
live_source=verified-shape-default
fi
test "${live_max}" = "${checked_max}"
cloud/dev/scripts/relay-asia-topology-workflow.test.mjs
+3
View File
@@ -90,6 +90,9 @@ test('checks the connection budget and production live ceiling before planning',
assert.match(workflow, /select\(\.name == "max_connections"\)/)
assert.match(workflow, /VERIFIED_DEFAULT_MAX_CONNECTIONS_TIER: db-custom-4-15360/)
assert.match(workflow, /VERIFIED_DEFAULT_MAX_CONNECTIONS_DATABASE_VERSION: POSTGRES_17/)
assert.match(workflow, /VERIFIED_DEFAULT_MAX_CONNECTIONS: '500'/)
assert.match(workflow, /live_max="\$\{VERIFIED_DEFAULT_MAX_CONNECTIONS\}"/)
assert.doesNotMatch(workflow, /live_max=\d/)
assert.match(workflow, /live_source=verified-shape-default/)
assert.match(workflow, /test "\$\(jq -er '\.settings\.tier'/)
assert.match(workflow, /test "\$\(jq -er '\.databaseVersion'/)
cloud/docs/relay-incident-monitor.md
+2 -1
View File
@@ -330,7 +330,8 @@ without its segment is a compile error in relay-contract, not a silent gap.
latest-sum over 24 healthy hours: mean ~100, 1-minute spikes to 216, with
10 minutes over the old bar of 160 — enough to freeze roughly one in ten
15-minute pre-drain gates on baseline noise. 250 cleared the healthy peaks
measured then and still fired well before the verified 400-connection ceiling;
measured then and still fired well before the 400-connection ceiling assumed at
the time (the live instance measured 500 on 2026-09-16);
pool waiters and pool wait latency keep their strict thresholds. Superseded by
the 2026-09-17 entry above, which re-measured a grown baseline against the
490-connection budget.